Highway Driving Safety Early Warning System Research Based On Monocular Vision

With the rapid development of economy, people’s living standards gradually improved andcars had entered innumerable families. Traffic safety problem gradually comes into the sight ofpeople, since the increase in car ownership. The highway, which regards as bridge connects witheach city，has become the place where traffic accident happened frequently. Highway safetyissues are an important part of intelligent transportation system. This paper is all about how toensure the security of high speed car.Paper research is based on monocular vision of highway driving safety early warning system.It is mainly improved the safety coefficient which driving fast in the highway. Follows are paper’smain content:(1) The introductory analysis and selection of image preprocessing methods are given in thispaper. Thinking of the demand for real-time analyses in the application and the effect after theimage processing. The model that base on monocular vision range and safety distance modelproposed by many researchers is analyzed in this paper.(2) Research on the particle filter algorithm and the TLD (Tracking-Learning-Detection)algorithm and compares each other advantage. In practical application, the particle filter algorithmcannot finish the target tracking reliably by itself, and can’t complete the task good enough in thehighway which have fast changing and complex background. TLD tracking algorithm have greateffect of tracking and can meet the need of user excellently, under the promise of real-time,because of its advantage: powerful tracking、learning and test mode.(3) Using data regression model to range, when part of camera parameters is unknown, makeuse of the relation between image and the mathematic of reality to establish mathematical model.First of all, this method is easy to implement in real life, research is more prefer to application;Secondly, a small amount of calculation, and if there is no guarantee that cannot be geared to theneeds of real-time applications for fast moving vehicles; Finally, the CCD camera focal length isunable to get in the practical application, the more advanced the camera you can get, the less forthe application.(4) The highway driving safety early warning system which combined TLD trackingalgorithm and the regression modeling ranging, gives tips in safe distance and. Can achieve thegoal of the auxiliary driving.